NOT FOUND ...' in conditional handled incorrectly".
Whenever we remove an instruction during optimization, we need to
adjust instruction numbers (ip - instruction pointer) stored in all
instructions. In addition to that, sp_instr_hpush_jump, which
corresponds to DECLARE CONTINUE HANDLER needs adjustment for m_handler,
which holds the number of instruction with the continue handler.
In the bug report, a wrong ip stored in m_handler was pointing at
FETCH, which resulted in an error message and abnormal SP termination.
The fix is to just remove m_handler member from sp_instr_hpush_jump,
as it's always points to the instruction next to the DECLARE
statement itself (m_ip+1).
and Bug#12297 SP crashes the server if data inserted inside a lon loop
Third commit attempt. With fixes to the issues, showed up after full rebuild and
tests on other hosts.
result set".
To enable full access to contents of I_S tables from stored functions
or statements that use them, we manipulate with thread's open tables
state and ensure that we won't cause deadlock when we open tables by
ignoring flushes and name-locks.
Building of contents of I_S.TABLES no longer requires locking of tables
since we use use handler::info() method with HA_STATUS_AUTO flag instead
of handler::update_auto_increment() for obtaining information about
auto-increment values. But this also means that handlers have to implement
support for HA_STATUS_AUTO flag (particularly InnoDB needs it).
* Don't activate prelocking mode for evaluating procedure arguments when it is not necessary.
* Code structure simplification and cleanup.
* Cleanup in .test files
its body, but lets each statement to get/release its own locks. This allows a broader set
of statements to be executed inside PROCEDUREs (but breaks replication)
This patch should fix BUG#8072, BUG#8766, BUG#9563, BUG#11126
crash if referencing a table" and several other related bugs.
Fix for bug #11834 "Re-execution of prepared statement with dropped function
crashes server." which was spotted during work on previous bugs.
Also couple of nice cleanups:
- Replaced two separate hashes for stored routines used by statement with one.
- Now instead of doing one pass through all routines used in statement for
caching them and then doing another pass for adding their tables to table
list, we do only one pass during which do both things.
We used 'IDENT' for labels as a temporary fix for the parser conflicts
introduced if the proper rule 'ident' was used. Now a specially tailored
'label_ident' rule is used for labels instead.
We want to have the defacto standard syntax for labels ("L:" instead of "label L;"),
and fix some known bugs, before we enable this again.
The code is left intact (#ifdef'ed SP_GOTO) and the test cases are kept in
sp-goto.test, for the future...
"Stored procedures: crash with function calling itself".
Disallow recursive stored routines until we either make Item's and LEX
reentrant safe or will use spearate sp_head instances (and thus separate
LEX objects and Item trees) for each routine invocation.
Fixed valgrind complaints. This fixes the memory leak problems for
procedured, and partially for functions. There's still a leak involving
results from functions that turned out to be too involved, so it will be
fixed separately.
Added a test case for bug #6866.
sql_select.cc:
Fixed bug #6866.
Bug was due to the fact that on_expr was not backed up
for the second execution of the stored procedure.
a stored procedure" (version 2).
To handle updates and inserts into view in SP properly we should set
lock types for tables of the view's main select when we are opening
view for prelocking purproses.
state" to sp-error.test.
According to Per-Erik all SP related tests which should result in error
should go into sp-error.test and not in sp.test, because we want to be
able to run sp.test using normal client.
illegal state".
We should not assume that mysql.proc table does not exist if we are
unable to open it under LOCK TABLES or in prelocked mode (and
remember this fact by setting mysql_proc_table_exists to zero).
and some SP-related cleanups.
- We don't have separate stage for calculation of list of tables
to be prelocked and doing implicit LOCK/UNLOCK any more.
Instead we calculate this list at open_tables() and do implicit
LOCK in lock_tables() (and UNLOCK in close_thread_tables()).
Also now we support cases when same table (with same alias) is
used several times in the same query in SP.
- Cleaned up execution of SP. Moved all common code which handles
LEX and does preparations before statement execution or complex
expression evaluation to auxilary sp_lex_keeper class. Now
all statements in SP (and corresponding instructions) that
evaluate expression which can contain subquery have their
own LEX.
and bug#8849 "problem with insert statement with table alias's":
make equality propagation work in stored procedures and prepared
statements.
Equality propagation can change AND/OR structure of ON expressions,
so the fix is to provide each execution of PS/SP with it's own
copy of AND/OR tree. We have been doing that already for WHERE clauses,
now ON clauses are also copied.
The warning sent is by itself ok, the problem was rather why it wasn't
sent on some other platforms...
The real problem was that a total_warn_count which was inconsistent with warn_list
was sent back with send_eof() during SP execution, which in turn cause a protocol
error in mysqltest.
BUG#6642: Stored procedure crash if expression with set function
BUG#7013: Stored procedure crash if group by ... with rollup
BUG#7743: 'Lost connection to MySQL server during query' on Stored Procedure
BUG#7992: SELECT .. INTO variable .. within Stored Procedure crashes
the server
BUG#8116: calling simple stored procedure twice in a row results
in server crash
Rolling back the item change list after each substatement in a procedure
fixed the failing assert().
Collect all tables and SPs refered by a statement, and open all tables
with an implicit LOCK TABLES. Do find things refered by triggers and views,
we open them first (and then repeat this until nothing new is found), before
doing the actual lock tables.
are not specified in an insert. Most of these changes are actually to
clean up the test suite to either specify defaults to avoid warnings,
or add the warnings to the results. Related to bug #5986.
This fixed a bug in prepared statements when used with outher joins
Fixed a bug in SUM(DISTINCT) when used with prepared statements.
Some safety fixes in test scripts to ensure that previous test failures shouldn't affect other tests
- No RESTICT|CASCADE in DROP SP (since it's not implemented)
- Added optional "noise" to FETCH: [[NEXT] FROM]
- At least one statement required in all block constructs except BEGIN-END
(where zero is allowed)
NO SQL
CONTAINS SQL (default)
READS SQL DATA
MODIFIES SQL DATA
These are needed as hints for the replication.
(Before this, we did have the default in the mysql.proc table, but no support in the parser.)
More tests.
Better error messages.
Fixed bug when checking if we updated all needed columns for INSERT.
Give an error if we encounter a wrong float value during parsing.
Don't print DEFAULT for columns without a default value in SHOW CREATE/SHOW FIELDS.
Fixed UPDATE IGNORE when using STRICT mode.
Under strict mode MySQL will generate an error message if there was any conversion when assigning data to a field.
Added checking of date/datetime fields.
If strict mode, give error if we have not given value to field without a default value (for INSERT)
Fixed (together with Guilhem) bugs in mysqlbinlog regarding --offset
Prefix addresses with 0x for easier comparisons of debug logs
Fixed problem where MySQL choosed index-read even if there would be a much better range on the same index
This fix changed some 'index' queries to 'range' queries in the test suite
Don't create 'dummy' WHERE clause for trivial WHERE clauses where we can remove the WHERE clause.
This fix removed of a lot of 'Using where' notes in the test suite.
Give NOTE instead of WARNING if table/function doesn't exists when using DROP IF EXISTS
Give NOTE instead of WARNING for safe field-type conversions
Althought techically not a but (as it's functioning as designed),
it was decided that the design should be changed. Some users have
a problem with dates being '0000-00-00' and the SQL standard specifies
that the modification date should be the same as the creation date
at creation.
The description is not entirerly correct. The issue was follow-up errors
where the first error is not caught - in which case it's often a system
error with errcode < 1000 (which are mapped by default to 'HY000'). In this
case the error state is different from what was assumed in the execution
loop.
at least partially. It doesn't crash or give packets out of order
any more, but it's unclear why it doesn't actually return anything
from within an SP. This should be investigated at some point, but
for the moment this will have to do. (It is a rather obscure feature... :)
Note: The following tests fails
- fulltext (Sergei has promised to fix)
- rpl_charset (Guilhem should fix)
- rpl_timezone (Dimitray has promised to fix)
Sanja needs to check out the calling of close_thread_tables() in sp_head.cc
It's not possible to quote the definition according to the current sql_mode
setting, so instead we use the setting stored with the SP (that's how it's
parsed anyway), and show this setting in the SHOW CREATE output.
BUG#1863: CREATE TABLE in Stored Procedure sometimes crashes on repeated calls.
BUG#2656: select with join in stored procedure: erroneous result on 2nd call.
BUG#3426: IF x IS NULL in stored procedure fails on second call within connection.
BUG#3448: Stored Procedures with inner joins possible bug.
BUG#3734: Stored procedure returns wrong rows with fulltext parameter.
BUG#3863: Stored procedure crash when incrementing variable in a loop.
(And corrected the row count output to the client after CALL)
Added new test cases for this, and adjusted old tests accordingly,
and new error codes and messages.
Fixed bugs in some tests (bug2673 and use test).
Added debug printing of instructions in SPs.
Procedure names were unintentionally case-sensitive when read from the database
(but case-insensitive when fetched from the cache).
Note that the DB-part of qualified names is still case-sensitive (for consistency
with other usage in mysql).
Phase 2: Make SPs belong to a DB, and use qualified names.
As a side effect, using USE in an SP is no longer allowed.
(It just doesn't work otherwise.)
